Reduction of interlaboratory variability in flow cytometric immunophenotyping by standardization of instrument set-up and calibration, and standard list mode data analysis.

Two workshops addressed the question to which degree standardization of instrument set-up and calibration, and standard list mode data analysis would reduce interlaboratory variability of flow cytometric results on prestained peripheral blood mononuclear cells (PBMC). Standard instrument set-up included uniform positioning of the "windows of analysis" for the forward and sideward light scatter and fluorescence (FL) 1 (i.e., fluorescein isothiocyanate [FITC]) and 2 (i.e., phycoerythrin [PE]) parameters. Reference standards and PBMC, double-stained with FITC- and PE-conjugated monoclonal antibodies covering a wide range of FL intensities and coexpression patterns, were sent out to 25 laboratories in Workshop 1 and to 35 laboratories in Workshop 2 with the following requests: a) to set up instruments according to local and standard protocols, b) to acquire list mode data on the PBMC with both instrument settings, and c) to analyze both datasets according to local protocols. Standard analysis of the list mode data acquired with uniform instrument settings was performed centrally using so-called "latent class model" software (Van Putten et al., Cytometry 14:86-96, 1993). This software provides an automated, "no-gating" analytical method of lymphocyte immunophenotypes and employs fixed FL marker settings as defined prior to each analytical run. In Workshop 1, these markers were set in identical histogram channels for all instruments based on results obtained with a reference instrument. Standard analysis of list mode data acquired after uniform instrument set-up led only to a 13% reduction of interlaboratory variability of results as compared to data analysis using local protocols. The standard protocol for instrument set-up led to uniform positioning of relatively strong FL signals but variable positioning of unstained cells on the FL histogram scales. Hence, standard FL marker settings were inappropriate for some instruments. Therefore, instrument responses to FITC and PE signals in Workshop 2 were calibrated using microbeads labeled with FITC or PE in a range of predefined FL intensities expressed in MESF units (molecules of equivalent soluble fluorochrome). That approach allowed the positioning of the FL markers for the standard analysis on the basis of identical FL1 and FL2 intensities, expressed in MESF units, for all instruments. Standard analysis of list mode data acquired after uniform instrument set-up and calibrated FL marker settings led to a 43% reduction of interlaboratory variability as compared to data analysis to local protocols. We conclude that standard list mode data analysis using fixed FL marker settings reduces the interlaboratory variability of flow cytometric results on prestained PBMC, provided that the instruments have been set up in a uniform way and that FL markers have been standardized on the basis of calibration of each instrument's response to the corresponding FL signals.